Introduction
In 2020 a global economic recession has begun, because of the coronavirus pandemic and subsequent government efforts to control it through lock-downs and similar measures. To help understand how economic contraction might change consumer behaviour we looked at what happened to consumer loyalty to brands in the last major recession (“the global financial crisis” or GFC).
Recessions are known for reducing total demand (indeed that is their technical definition), so consumer behaviour is clearly changing. But aside from less spending overall, how do consumers react? Some pundits suggest a collapse in brand loyalty is likely, along with a temporary or even permanent flight to cheaper brands (or store brands in the case of grocery). While this sounds intuitive, it could also be the case that in tough times consumers prefer to stick to brands they know.
We start with some very recent analysis of supermarket shopping behaviour during the start of the Coronavirus (COVID-19) pandemic. We then turn to our central topic of economic recession. We discuss durables and services and then turn more specifically to consumer packaged goods, an area which has been more heavily researched. We address two broad questions:
- How does shopper behaviour change in a recession?
- How should consumer brands market in a recession?
How did shoppers behave during the pandemic?
First, we answer one question Institute sponsors have asked about the current Coronavirus situation: have shopping habits changed? We obtained Australian shopping data for three two-week periods: a baseline before Coronavirus (3/2/20-16/2/20), during the peak of panic buying, before Coronavirus arrived on our shores (9/3/20-22/3/20), and during full lockdown (27/4/20-10/5/20), where people were told to stay home unless absolutely necessary, many workplaces and non-essential stores were shut, and so on.

The most important message to take away is that the patterns of buying behaviour described in Report #64 are incredibly robust. This can be seen, by comparing the three time periods on the charts below:
Items per Basket by % of Visits

Items per Basket by n of Visits

Spend Break by % of Visits

Panic buying is real, but still follows established patterns
Panic buying was, of course, a real phenomenon – in aggregate. During the stockpiling/panic buying period, the total number of store visits went up by 17% (an additional 5.7m visits), $ sales increased by 34% (over $400m), and units sold also increased by 34% (110m units). Baskets over $55 AUD in value increased by 38% (from 21% to 25% of all baskets), and more astonishingly, baskets over $200 AUD in value increased by 71% (from 2.1% to 3.1% of all baskets). You can see this in the chart of absolute number of shopping baskets per size: the number of trips for each size of basket grew across the distribution.
While this behaviour is newsworthy, the fundamental behaviour – namely, many small baskets – remained the same: a shift in basket sizes of only 2.2 percentage points away from baskets under ten items to baskets over ten items was enough to result in shortages. But, we still saw the same fundamental behaviour of one in five visits being for a single item, and half of all visits being for fewer than five items. Reality contrasts with newspaper stories of trolleys full of toilet paper, rice and pasta!
Normal shopping behaviour returns during lockdown
When lockdown was introduced, shopping behaviour returned, in some ways, almost to normal: while the overall volume of shopping trips dropped by 15% from baseline (a drop of 5.2m trips), $ sales only decreased by 0.7% (compared to baseline; -5.5% by units). The chart of absolute numbers shows that the volume of trips dropped below the baseline by quite a bit. But, the powerful needs people have for grocery items can’t be easily ignored.
As expected in the midst of a pandemic, people weren’t making as quite as many trips out of the house for a single item. Though this change was small. When things are normal, 21% of all shopping trips are for a single item. This barely changed during panic buying, but did change slightly during lockdown, to “only” 18%.
Small baskets continue to dominate
Another way to look at this preponderance of small baskets is by considering the half-way point of the data (the median): half of all visits before Coronavirus are for baskets of less than four items. This accounts for 20% of sales, and 13% of units purchased. During peak panic buying, half (49%) of all baskets are still for four items or fewer, which is 17% of sales, and 12% of units purchased. During lockdown, this still didn’t change all that much: half of all visits (52%) are for fewer than five items, which accounts for 21% of sales, and 14% of units purchased. One measure that did change is the overall dollar value of baskets: from 50% of total sales coming from baskets <$75 at baseline, to a peak of <$90 during panic buying, and back down to <$85, this, combined with the -5.5% drop in unit sales noted above, suggests that shoppers are perhaps buying larger pack sizes to account for the decreased total number of trips.
Some more visitation of small stores
We saw a decrease in market share of the two largest grocery chains in Australia, and an increase in share for the smaller supermarkets. This might be due to government-mandated restrictions to the number of people who could enter a store. This led to queues outside stores, and so shoppers visited other, small ones. It might also be due to retailer’s online services being suddenly overwhelmed with demand (a jump from 6% to 8% of total buying in two months), leading them to restrict their service to vulnerable groups only. This led to customers visiting a slightly wider repertoire of retailers to get products they couldn’t get from the main players. While it’s possible that this will return to normal share after all restrictions are relaxed, it is incumbent upon the beneficiaries of these changes to make sure that they provide a competitive offering, and become part of the repertoire of customers.
When asked “What will happen when things return to normal?”, perhaps the answer is that in many ways, they already have – even in lockdown. Despite the huge differences in the total number of trips made, we don’t see much difference in the overall pattern of shoppers making lots and lots of very small trips in either panic buying stage or lockdown. Our advice to retailers: that they should cater for this great number of small trips remains unchanged, and our advice to manufacturers: that they need to be aware of the fleeting attention spans they’ll receive in stores, and to ensure they are well branded, using strong Distinctive Assets across all touchpoints, also remains unchanged.
We now turn to discuss what happens in economic recessions.
Different sectors suffer differently
Services
There is surprisingly little evidence about what happens to service brands in recessions. By definition, if the economy contracts that means less is being spent on goods and services economy-wide, so we would naturally expect the services sector to contract in a recession. But it might not be particularly sensitive. For example, an analysis of over 50,000 US hotels found, perhaps surprisingly, only marginally lower hotel occupancy rates in recession years of 2002 and 2008 compared to the intervening period. Average prices did not fall in those recession years either (O’Neill & Carlbäck, 2011). Of course, in today’s pandemic, hotels, restaurants, bars, and other services associated with hospitality and tourism have been especially heavily hit in most countries. The recovery for these services is highly dependent on the re-emergence of a travelling public. We do not anticipate this to be particularly rapid. Firstly, many countries will maintain social distancing for some period of time and secondly, international travel has largely ceased. It will be some time before governments relax restrictions and the public being willing to travel again, or even mingle to the same extent.
For large services firms such as banks, this recession is obviously not good news, but it is not like the 2008 GFC. In 2008, there was a crisis of confidence within the financial services sector. A number of banks had to be saved by government intervention, and others came close to failing. In 2020, stock prices for leading banks in Europe, Australasia and North America have all dropped around 30% in anticipation of lower profits. However, stock prices are recovering, and will probably continue to do so as economies gather pace. The principal business challenge for banks now, as with all services companies, is to maintain and win market share, as it always has been. Interestingly in Australia, banks have substantially increased their advertising spend in recent months.
Durables
Durables are considered to be more sensitive to the economy than packaged goods. Consumers are more able to postpone buying durables in hard times. We can put up with an old washing machine or refrigerator for another year if we feel we have to. By contrast, many packaged goods such as food, drink and personal care items are less easily gone without. A wide-ranging study of durables demand in the US found that sales of durables are particularly sensitive to the business cycle. The study reports quite sharp sales drops for 25 classes of durables when overall economic growth is low. Moreover, it found that while the demand drops are quick, sales take years to climb back to former levels (Deleersnyder, Dekimpe, Sarvary & Parker, 2004). Of course, durable purchases cannot always be postponed indefinitely. Nonetheless, the immediate effect of a recession can be substantial – new car sales in Australia were down 18% in March (YoY) – though patchy, for example Toyota grew sales by 1.6%. In the US, car sales in 2020 are down around 20% compared to 2019 (Thiel, 2020).
Consumer Packaged goods
There is much more evidence about what happens in packaged-goods markets in recessions given the huge amount of purchasing data tracked in this sector.
We now address two broad questions:
- How does shopper behaviour change in a recession?
- How should consumer good brands market in a recession?
Do consumers turn to Private Label Brands?
Store Brands reportedly do better in hard economic times. In the 1990’s, US researchers concluded that store brand market share was higher when disposable income growth was lower (Hoch & Banerji, 1993). They did caution, though that some of the effect may be due to national brands reducing advertising:
“Manufacturers often set advertising budgets as a percentage of sales, so that they decrease their brand support during economic downturns, exacerbating the consumer income effect. Retailers, anticipating this consumer income effect, typically increase investment in private labels during downturns” (p.58)
A more recent study examined the gains and losses of share for Store Brands over a span of about 25 years in the US, UK, Germany and Belgium (Lamey, Deleersnyder, Dekimpe & Steenkamp, 2007), i.e. from the 1970’s-80’s to the early 2000’s. The key finding was that store brands tended to grow share a tiny bit faster during a recession. Again, the authors pointed out that national brands tended to drop their advertising budgets and new product launches in low-GDP growth time periods (consistent with Hoch & Banerji), suggesting
“their behaviour might actually exacerbate the negative effects of an economic downturn
on their brand’s positions” (p.11)
While we do have some evidence that hard times are advantageous for store brands, the losses for national brands could be at least partially self-inflicted.
We checked what happened with Private Label (PL) share in the last recession. We found PLs did increase share in some categories, but lost share in others. As the table below shows, PLs (as a whole) lost share in 4 of the 9 categories we had available data to analyse. (Note we do not have the PL data for dog food, which appears in some of the other analyses).
Table 1: Store Brand share over time, UK

Do consumers become more price promotion sensitive?
Next, we examined the incidence of buying on price-promotion, i.e. the change in category volume that is sold on temporary discounts. Intuitively, we might expect that in an economic downturn, manufacturers (and retailers) might put their brands on deal for more weeks of the year. Furthermore, consumers may pay more attention to price specials and thus increase the extent to which they buy on deal. We show results for the UK below. Here we do see a change over time – the proportion of sales sold on promotion increases from 34 to 37% in 2008. Notably, however, it increases by about similar amounts over the following years. Therefore, the 2008 increase may not be solely due to the recession, but could just be part of an increase in price promotions over time.
Table 2: Promotional activity over time, UK

Did consumers buy cheaper?
So, there was an increase in buying on discount during the last recession, but did this mean that consumers overall bought more cheaply? It seems a reasonable thing to expect, not only due to buying on discount but also consumers may ‘trade down’ to cheaper brands due to tighter budgets. But surprisingly, we do not see this effect in the data that we have for the UK. In Table 3 we show average price-paid-per-unit-of-weight for brands in the same ten UK categories, for 2007-2011. As we see in the table, average prices stay very stable.
Table 3: Average price over time

Given there was a small increase in buying on-deal the above table does not show consumers trading down to cheaper brands. Undoubtedly some households did, but they must have been matched by some households that traded up.
Does Brand Loyalty falter in recessions?
Intuitively we might think that a contraction in spending and employment and lower consumer confidence might induce shoppers to make changes; thus eroding brand loyalty. They might seek out lower prices, or they might seek to treat themselves by buying better quality. Indeed, consumers might do both at different times and in different categories. All of which may disrupt loyalties, resulting in consumers having larger repertoires.
Counter to that view is research that has examined the stability of brand loyalty over long periods. A study by Ehrenberg-Bass researchers across numerous US and UK categories for the period 1998-2010 (Dawes, Meyer-Waarden & Driesener, 2015) found considerable stability in loyalty over the period – which included the great recession of 2008. This stability suggests the recession did not cause consumers to dramatically expand their repertoires e.g. searching for better value.
We can also directly examine the evidence of what happened during the last recessionary period, in 2008. Using data on consumer purchases in the years just before and after 2008 (in the UK), and 2009 (in the USA) we looked at whether brand loyalty dropped as a consequence of the recession. We found that loyalty did not erode.
United States
Using data provided by Nielsen we tabulated average brand Share of Category Requirements (SCR) for brands in a dozen USA CPG categories from 2006-2010. What we see is perhaps surprising – essentially no change in brand loyalty in the period of the recession compared to the immediate years before or after. This strongly suggests consumers did not markedly change their purchasing behaviour in relation to grocery brands by, for example broadening their repertoires looking for cheaper alternatives. If that had been the case, brand SCR would have dropped, but it did not.
To explain the data further, SCR refers to how much of a consumer’s category purchasing a given brand satisfies for its buyers. So, if the buyers of Pantene buy it half the time that they purchase shampoo then Pantene has an SCR of 50%. In Table 4 we report on the average for all of the major brands in each category. A high SCR for the category means that buyers have small repertoires, due to high loyalty and also the category purchase rate, i.e. repertoires are smaller when people don’t buy often. We can see that average SCR varies between categories, although here most are between 20-40%. Deodorants are noticeably higher, which is understandable given they are infrequently purchased. In contrast, buyers of ready-to-eat breakfast cereal have large repertoires, reflecting higher purchase frequency but also suggesting that variety seeking is a marked aspect of this category.
Interestingly, pet food categories show considerable differences in repertoire size, repertoires are smaller for dog food buyers than cat food buyers (perhaps cats really are fussy eaters), and smaller for buyers of dry than wet pet food (again probably reflecting a purchase frequency differences).
Table 4: Loyalty over time US

United Kingdom
We first tabulated average repertoire sizes (the average number of brands consumers purchased over 12 months) for a range of UK CPG categories for the year 2007-2011. As with the US data we see no meaningful change in repertoires in 2008. Across the ten categories, average repertoire stayed essentially the same over the five-year period 2007-2011, at around 3.8 brands.
Table 5: Repertoire size over time

This strongly supports the US finding that loyalty (in so far as it is measured by SCR and repertoire size respectively) remains remarkably stable at the aggregate level. Which, in turn, suggests that a change in the nature of consumers is not the problem.
This does not mean that all brand shares were stable, they weren’t. But there was a high degree of stability.
We now examine the question of how marketing activity works in times of contraction, and whether basic patterns in brand performance still hold.
How should consumer goods brands market in a recession?
Do marketers need to invest more during recessions?
Quite a number of pundits have strongly argued that a recession is the time to invest more into marketing activity, especially advertising. However, much of the research in this area comes from media companies with a vested interest in encouraging advertising spend.
It would appear then that unless there is an increase in competitor marketing activity, there is no convincing evidence marketers need to invest more marketing dollars in recessions. Rather the focus should be on maintaining spend in the face of competitor reductions.
Investment in advertising generally tends to drop in recessions. A large study by Picard (2001) across nine countries found advertising spending reduced by about 5% for every one percentage point drop in GDP. Therefore, even just maintaining the pre-recession level of advertising will very likely increase a brand’s share of voice as aggregate ad spend decreases. And evidence suggests that if share of voice goes up over a sustained period such as a year or more, there is a tendency (not guaranteed!) to also pick up some market share. As we reported in a previous point, researchers have long suspected that any pick-up in Private Label share in recessions was at least partially because National Brands tended to drop their advertising support and new product activity in recessions. As a consequence, collective spend for National Brands was lower, allowing Private Labels to gain market share.
Should we focus on our most loyal customers?
The Ehrenberg-Bass Institute is often asked in normal times (let alone during a pandemic) whether a brand should focus its marketing efforts on its most loyal buyers. The idea of “hunkering down” during a recession, i.e. reducing reach and focusing on the brand’s most loyal/heavy buyers, appears to be intuitively appealing for many marketers.
To respond to this question, we looked at stable brands, as well as brands that grew or declined from 2007 to 2008 by correlating relative change in market share, and relative change in penetration and purchase frequency. The data clearly shows that market share growth or decline during a recession is strongly explained by penetration growth or decline (see Figure 1). That is, in line with the Double Jeopardy law, brands that maintained their market share in a recession did so by holding onto their customers, especially their light customers. Brands that grew market share did so by growing their customer base, and brands that declined lost customers.
Figure 1: Relationship between change in penetration and change in market share

In line with Double Jeopardy, growing brands had a (slight) tendency to also grow purchase frequency, but it is remarkable how weakly related purchase frequency changes were to growth or decline in the recession (see Figure 2).
Figure 2: Relationship between change in loyalty and change in market share

To understand more about brand growth or decline, it is useful to deconstruct the brand’s buyer base over two time periods. Data of this nature is harder to come by, nevertheless we present two examples here. One is for a brand that grew, and the other is a brand that declined.
We firstly show the distribution of non, light, medium and heavy buyers (i.e. non buyers, to the brand’s most loyal buyers) in two time periods. In charts 1,2,3, we show the brand that grew in the second year– it gained more of every buyer type, but note the marked change in light buyers.
Charts 1,2,3: Change in buyer groups – growing brand



There is change over all buyer classes, however most change occurs at the light buyer level.
In charts 4,5,6 below, the brand declines – it lost light, medium and heavy buyers but the very big drop-off is in light buyers.
Charts 4,5,6: Change in buyer groups – declining brand

There is change over all buyer classes, however most change occurs at the light buyer level.
The clear implication is that growth or decline – whether it be in a recession or in ‘normal’ times – involves changes among all buyer groups, although the big changes appear in the light buyers. The evidence simply does not support the intuitive idea that, during a recession, loyal / heavy buyers should receive a higher than normal allocation of effort and resources.
Does brand growth associate with price cut or promotion?
To answer this question, we looked the relative change in market share, and the relative change in price and % sold on promotion from 2007 to 2008. In fact the changes in price we see here are mostly a result of altering promotions – doing fewer promotions pushes the brand’s average price up a bit; doing more depresses price downwards.
The results show that growth and decline during recession has little association with price cut or price promotion. This is in line with our findings (forthcoming) on long-term brand growth: changes in price levels and promotion over the course of one year to another have little effect on brand growth.
It is important to contextualise this finding, which might be surprising. We do know price promotions have strong, short-term effects on brand sales. And we do know that price elasticity is, on average -2.5. But the effects of price promotions are seen very apparently in short periods such as a week – sales spike up then snap back to normal. If we look at changes in price promotion for a brand over longer periods – say, a brand goes from 30% sold on deal in one year to either 35% or 25% in the next year – we see much weaker associations with change in sales or market share. This is because a lot of the immediate sales effects are forward-buying (and in many cases cannibalising from the rest of the range). When we aggregate to longer time periods the sales gains from any particular week’s promotions to the brand’s cumulative sales are dampened down. That partly explains the relatively low association in the two charts.
The other big factor that likely explains why we see a weak association here is because of the reasons why brand managers decide to change their promotions regime, or make significant changes to normal selling price. And that is, brands that are in trouble are more likely to escalate promos or drop price; growing ones are less likely to. And what these charts communicate is, that those things don’t tend to change the brand’s trajectory.
We do know the big factors that shape a brand’s market share are mental and physical availability. If those things don’t change, making changes to price or the incidence of promotion from one year to another may not have much of an effect on brand success.
Figure 3: Relationship between change in price and change in market share

Figure 4: Relationship between change in promotion and change in market share

In summary, we find that:
Consumer loyalty did not appear to change in the course of the last recession.
Private Label / Store Brand market share did not generally increase, at least in the data we were able to analyse. In some categories Store Brands grew, but in some others they declined.
Prices paid by consumers remained quite stable at the category level across a number of categories. With no evidence of any ‘trading down’ trend.
While buying on price promotions increased slightly, the rise could simply be part of a longer-term upward trend that continued after the recession.
The fact that repertoire sizes stayed the same suggests that the small increase in promotion buying was almost entirely consumers spotting a brand already in their repertoire that is on deal, rather than searching and trying previously-untried brands that happen to be on promotion.
Brand growth/decline in market share was mainly associated with growth/decline in penetration and slightly associated with loyalty – following the DJ pattern.
Brand growth/decline in market share was not associated with price cut or promotion.
Stocks/shares in consumer goods companies are widely considered recession proof, along with shares in healthcare, alcohol (but perhaps not Champagne), and for retailers of these products. Our findings support this view.
Shoppers continue to make mostly many small trips to supermarkets, but might be buying larger pack sizes to account for an overall decrease in the number of trips.
Conclusion:
While no recession is completely the same, our analysis makes us confident that the current evidence-based rules for brand growth still apply in a recessionary period. Marketing strategy therefore still needs to focus on maintaining (ideally building) mental and physical availability.
We strongly urge against knee-jerk reactions to either the pandemic or the recession. There is good anecdotal evidence that much of the pandemic inspired advertising (ie attempts to reassure customers) was not needed, and/or resulted in bland lookalike advertising that failed to build distinctive assets or mental availability.
Given that price promotions have declined in recent months we encourage our sponsors to ask whether or not they should return to higher levels of promotion. Examine what has happened to profitability during this period.
Our top recommendation is to avoid listening to marketing gossip, stay sceptical, and carefully collect evidence to guide marketing investments. Find out what really is changing in the world (eg shifts to new media, new sales channels), and how much. Invest in evidence in order to make better safer marketing decisions.